In recent years, continuum reverberation mapping involving high cadence UV/optical monitoring campaigns of nearby Active Galactic Nuclei has been used to infer the size of their accretion disks. One of the main results from these campaigns has been that in many cases the accretion disks appear too large, by a factor of 2 - 3, compared to standard models. Part of this may be due to diffuse continuum emission from the broad line region (BLR), which is indicated by excess lags around the Balmer jump. Standard cross correlation lag analysis techniques are usually used to just recover the peak or centroid lag and can not easily distinguish between reprocessing from the disk and BLR. However, frequency-resolved lag analysis, where the lag is determined at each Fourier frequency, has the potential to separate out reprocessing on different size scales. Here we present simulations to demonstrate the potential of this method and then apply a maximum likelihood approach to determine frequency-resolved lags in NGC 5548. We find that the lags in NGC 5548 generally decrease smoothly with increasing frequency, and are not easily described by accretion disk reprocessing alone. The standard cross correlation lags are consistent with lags at frequencies lower than 0.1 per day, indicating they are dominated from reprocessing at size scales greater than about 10 light days. A combination of a more distant reprocessor, consistent with the BLR, along with a standard-sized accretion disk is more consistent with the observed lags than a larger disk alone.